[Serving] add ppdet serving example (#641)

* serving support ppdet

* Update README.md

update ppadet/README

examples/vision/detection/paddledetection/serving/models/runtime/1/README.md

@@ -0,0 +1,5 @@
+# Runtime Directory
+
+This directory holds the model files.
+Paddle models must be model.pdmodel and model.pdiparams files.
+ONNX models must be model.onnx files.

examples/vision/detection/paddledetection/serving/models/runtime/faster_rcnn_runtime_config.pbtxt

@@ -0,0 +1,58 @@
+backend: "fastdeploy"
+
+# Input configuration of the model
+input [
+  {
+    # input name
+    name: "image"
+    # input type such as TYPE_FP32、TYPE_UINT8、TYPE_INT8、TYPE_INT16、TYPE_INT32、TYPE_INT64、TYPE_FP16、TYPE_STRING
+    data_type: TYPE_FP32
+    # input shape， The batch dimension is omitted and the actual shape is [batch, c, h, w]
+    dims: [ -1, 3, -1, -1 ]
+  },
+  {
+    name: "scale_factor"
+    data_type: TYPE_FP32
+    dims: [ -1, 2 ]
+  },
+  {
+    name: "im_shape"
+    data_type: TYPE_FP32
+    dims: [ -1, 2 ]
+  }
+]
+
+# The output of the model is configured in the same format as the input
+output [
+  {
+    name: "concat_12.tmp_0"
+    data_type: TYPE_FP32
+    dims: [ -1, 6 ]
+  },
+  {
+    name: "concat_8.tmp_0"
+    data_type: TYPE_INT32
+    dims: [ -1 ]
+  }
+]
+
+# Number of instances of the model
+instance_group [
+  {
+    # The number of instances is 1
+    count: 1
+    # Use GPU, CPU inference option is:KIND_CPU
+    kind: KIND_GPU
+    # The instance is deployed on the 0th GPU card
+    gpus: [0]
+  }
+]
+
+optimization {
+  execution_accelerators {
+  gpu_execution_accelerator : [ {
+    # use Paddle engine
+    name: "paddle",
+  }
+  ]
+}}

examples/vision/detection/paddledetection/serving/models/runtime/mask_rcnn_runtime_config.pbtxt

@@ -0,0 +1,63 @@
+backend: "fastdeploy"
+
+# Input configuration of the model
+input [
+  {
+    # input name
+    name: "image"
+    # input type such as TYPE_FP32、TYPE_UINT8、TYPE_INT8、TYPE_INT16、TYPE_INT32、TYPE_INT64、TYPE_FP16、TYPE_STRING
+    data_type: TYPE_FP32
+    # input shape， The batch dimension is omitted and the actual shape is [batch, c, h, w]
+    dims: [ -1, 3, -1, -1 ]
+  },
+  {
+    name: "scale_factor"
+    data_type: TYPE_FP32
+    dims: [ -1, 2 ]
+  },
+  {
+    name: "im_shape"
+    data_type: TYPE_FP32
+    dims: [ -1, 2 ]
+  }
+]
+
+# The output of the model is configured in the same format as the input
+output [
+  {
+    name: "concat_9.tmp_0"
+    data_type: TYPE_FP32
+    dims: [ -1, 6 ]
+  },
+  {
+    name: "concat_5.tmp_0"
+    data_type: TYPE_INT32
+    dims: [ -1 ]
+  },
+  {
+    name: "tmp_109"
+    data_type: TYPE_INT32
+    dims: [ -1, -1, -1 ]
+  }
+]
+
+# Number of instances of the model
+instance_group [
+  {
+    # The number of instances is 1
+    count: 1
+    # Use GPU, CPU inference option is:KIND_CPU
+    kind: KIND_GPU
+    # The instance is deployed on the 0th GPU card
+    gpus: [0]
+  }
+]
+
+optimization {
+  execution_accelerators {
+  gpu_execution_accelerator : [ {
+    # use Paddle engine
+    name: "paddle",
+  }
+  ]
+}}

examples/vision/detection/paddledetection/serving/models/runtime/ppyolo_runtime_config.pbtxt

@@ -0,0 +1,58 @@
+backend: "fastdeploy"
+
+# Input configuration of the model
+input [
+  {
+    # input name
+    name: "image"
+    # input type such as TYPE_FP32、TYPE_UINT8、TYPE_INT8、TYPE_INT16、TYPE_INT32、TYPE_INT64、TYPE_FP16、TYPE_STRING
+    data_type: TYPE_FP32
+    # input shape， The batch dimension is omitted and the actual shape is [batch, c, h, w]
+    dims: [ -1, 3, -1, -1 ]
+  },
+  {
+    name: "scale_factor"
+    data_type: TYPE_FP32
+    dims: [ -1, 2 ]
+  },
+  {
+    name: "im_shape"
+    data_type: TYPE_FP32
+    dims: [ -1, 2 ]
+  }
+]
+
+# The output of the model is configured in the same format as the input
+output [
+  {
+    name: "matrix_nms_0.tmp_0"
+    data_type: TYPE_FP32
+    dims: [ -1, 6 ]
+  },
+  {
+    name: "matrix_nms_0.tmp_2"
+    data_type: TYPE_INT32
+    dims: [ -1 ]
+  }
+]
+
+# Number of instances of the model
+instance_group [
+  {
+    # The number of instances is 1
+    count: 1
+    # Use GPU, CPU inference option is:KIND_CPU
+    kind: KIND_GPU
+    # The instance is deployed on the 0th GPU card
+    gpus: [0]
+  }
+]
+
+optimization {
+  execution_accelerators {
+  gpu_execution_accelerator : [ {
+    # use Paddle engine
+    name: "paddle",
+  }
+  ]
+}}

examples/vision/detection/paddledetection/serving/models/runtime/ppyoloe_runtime_config.pbtxt

@@ -0,0 +1,55 @@
+# optional, If name is specified it must match the name of the model repository directory containing the model.
+name: "ppyoloe_runtime"
+backend: "fastdeploy"
+
+# Input configuration of the model
+input [
+  {
+    # input name
+    name: "image"
+    # input type such as TYPE_FP32、TYPE_UINT8、TYPE_INT8、TYPE_INT16、TYPE_INT32、TYPE_INT64、TYPE_FP16、TYPE_STRING
+    data_type: TYPE_FP32
+    # input shape， The batch dimension is omitted and the actual shape is [batch, c, h, w]
+    dims: [ -1, 3, -1, -1 ]
+  },
+  {
+    name: "scale_factor"
+    data_type: TYPE_FP32
+    dims: [ -1, 2 ]
+  }
+]
+
+# The output of the model is configured in the same format as the input
+output [
+  {
+    name: "multiclass_nms3_0.tmp_0"
+    data_type: TYPE_FP32
+    dims: [ -1, 6 ]
+  },
+  {
+    name: "multiclass_nms3_0.tmp_2"
+    data_type: TYPE_INT32
+    dims: [ -1 ]
+  }
+]
+
+# Number of instances of the model
+instance_group [
+  {
+    # The number of instances is 1
+    count: 1
+    # Use GPU, CPU inference option is:KIND_CPU
+    kind: KIND_GPU
+    # The instance is deployed on the 0th GPU card
+    gpus: [0]
+  }
+]
+
+optimization {
+  execution_accelerators {
+  gpu_execution_accelerator : [ {
+    # use Paddle engine
+    name: "paddle",
+  }
+  ]
+}}